Are carbon films worth the trouble?

[bepone]

Why does that matter?

for you then is the same, if someone wrote this is noisy resistor, then you don't care for that material, because it is the same from 10 ohms to 1000000 ohms all are noisy on the same level

[WhopperPlate]

Why does that matter?

Sounds like one more reason to consider alternatives within the power supply If mox are noisier within any given application .

[nworbetan]

MOX are noisy. I still wonder if this resistor mojo isn't just about noise.

I remember someone here once posted an article discussing how resistor noise significantly contributes to the tonal nuances of each resistor . I can’t cite it or find it but I remember reading about it all

I can't say if this is the same article, but do I have this one page about resistor noise (a section of a larger text about resistors in general) bookmarked.

https://eepower.com/resistor-guide/resi ... tor-noise/#

[R.G.]

Why does that matter?

Sounds like one more reason to consider alternatives within the power supply If mox are noisier within any given application .

Right up until you notice that in power supplies, the resistors are bypassed to ground with BFCs (Big Freaking Caps). The noise is reduced by the filtering effect of the caps and even more so by later-in-the-chain resistors and bypass caps, which further reduce the noise, as well as any common mode rejection of circuit stages. Even the full thermal-plus-excess noise of a resistor in the power supply is tiny compared to the power supply voltage itself.

Even the use of carbon comps in power supplies doesn't make the power supplies noticeably noiser, and they were used there in lots of vintage circuits when their power rating was enough. MOX, being lower noise, even if not perfect, also would make little difference to a power supply output.

Questions like this are where being able to express things in numbers is a huge advantage. Ever lower noise resistors only matter when their noise voltage/current outputs are large enough to matter compared to the circuit voltages. I did a quick calculate of the thermal noise of a 10K resistor (only the resistance and temperature matter for thermal noise) at 25C. It came out to 1.283 microvolts. Compared to a 400V B+, that's -169.9db. But there is also excess noise, where CC has problems. Excess noise is current noise, which varies as 1/F. At 82 Hz (excess is worse at lower frequencies) a 10K CC resistor has an excess noise of (as nearly as I can squint the graph) 23db more than a MOX. MOX is about 6db higher than wirewound, which is substantially the same as its thermal noise. (https://pub.dega-akustik.de/ICA2019/dat ... 001261.pdf)
So we could estimate that a 10K MOX would have -164db thermal plus excess noise at 82Hz, tapering to -170db by ~400hz.

Why would we worry about MOX noise in power supplies?

[WhopperPlate]

Why does that matter?

Sounds like one more reason to consider alternatives within the power supply If mox are noisier within any given application .

Right up until you notice that in power supplies, the resistors are bypassed to ground with BFCs (Big Freaking Caps). The noise is reduced by the filtering effect of the caps and even more so by later-in-the-chain resistors and bypass caps, which further reduce the noise, as well as any common mode rejection of circuit stages. Even the full thermal-plus-excess noise of a resistor in the power supply is tiny compared to the power supply voltage itself.

Even the use of carbon comps in power supplies doesn't make the power supplies noticeably noiser, and they were used there in lots of vintage circuits when their power rating was enough. MOX, being lower noise, even if not perfect, also would make little difference to a power supply output.

Questions like this are where being able to express things in numbers is a huge advantage. Ever lower noise resistors only matter when their noise voltage/current outputs are large enough to matter compared to the circuit voltages. I did a quick calculate of the thermal noise of a 10K resistor (only the resistance and temperature matter for thermal noise) at 25C. It came out to 1.283 microvolts. Compared to a 400V B+, that's -169.9db. But there is also excess noise, where CC has problems. Excess noise is current noise, which varies as 1/F. At 82 Hz (excess is worse at lower frequencies) a 10K CC resistor has an excess noise of (as nearly as I can squint the graph) 23db more than a MOX. MOX is about 6db higher than wirewound, which is substantially the same as its thermal noise. (https://pub.dega-akustik.de/ICA2019/dat ... 001261.pdf)
So we could estimate that a 10K MOX would have -164db thermal plus excess noise at 82Hz, tapering to -170db by ~400hz.

Why would we worry about MOX noise in power supplies?

Not so much worry , just one more consideration in the sum of the parts. Thanks for the technical demo .

[nworbetan]

Questions like this are where being able to express things in numbers is a huge advantage. Ever lower noise resistors only matter when their noise voltage/current outputs are large enough to matter compared to the circuit voltages. I did a quick calculate of the thermal noise of a 10K resistor (only the resistance and temperature matter for thermal noise) at 25C. It came out to 1.283 microvolts. Compared to a 400V B+, that's -169.9db.

I'm not familiar with calculating a signal to noise ratio using a DC power supply voltage as the "signal" that the noise is relative to. Wouldn't it be more appropriate to do a S/N calculation using the ripple voltage that rides on top of the 400V B+?

Or, even better, what would the S/N ratio be with a 100k resistor on the plate of a triode? That seems like some math that would be pretty appropriate considering the source you linked is primarily focused on the passive intermodulation (PIM) that distorts an actual signal by intermodulating it with noise.

[dorrisant]

Food for thought or argument...

https://groups.google.com/g/rec.audio.t ... 9iBk?pli=1

[Reeltarded]

Tube amps are not ideal circuits. What perfection do we strive for? For most of my life I have been willing to play a waterfall of hiss louder than many bands and control the breakdown of reality with my pinky on a volume knob. Do you have any idea what a Super Hard On thru an AC30 does? Yikes.

*imagine your band owns a 707 and it's parked in your house and kept running*

We should really never argue about the small bits. We should argue about something that REALLY matters.

Output Transformers

That said, MOX certainly works in power strings, is easy to find in useful values, and I don't use it in other places because I don't plan on buying those values in MOX.

PS: Does anyone have a pile of Piher 2k, 5k, 10k, 20k and 56k in 2w-5w for sale? Iskras might do.

[bepone]

2W iskra cf: 1k,2k, 3k, 4k7, 10k, 27k, 68k ..

[WhopperPlate]

Tube amps are not ideal circuits. What perfection do we strive for? For most of my life I have been willing to play a waterfall of hiss louder than many bands and control the breakdown of reality with my pinky on a volume knob. Do you have any idea what a Super Hard On thru an AC30 does? Yikes.

*imagine your band owns a 707 and it's parked in your house and kept running*

I call that “riding the elephant”… some people look at you when you hit the standby switch like you shot their dog …

2W iskra cf: 1k,2k, 3k, 4k7, 10k, 27k, 68k ..

Ah cmon , quit teasing me

I have some 470 ohm iskra I am gonna try eventually

I got 10k and 8.2k . Marshall lover at heart obviously

Fwiw Koa Speer and yageo still produce higher wattage cf in many values

[R.G.]

I'm not familiar with calculating a signal to noise ratio using a DC power supply voltage as the "signal" that the noise is relative to. Wouldn't it be more appropriate to do a S/N calculation using the ripple voltage that rides on top of the 400V B+?

I'm not familiar with calculating signal to noise on a DC power supply output either. I had always just assumed that resistor noise in a power supply could be ignored as too small to matter on the power supply output. I did the math to sound out my internal assumptions. In this one instance, my assumptions were correct: even if you ignore noise reduction by capacitor filtering, the noise of a MOX resistor is incredibly trivial in a power supply. Even an extra-noisy CC's noise is too small to worry about in a power supply, by the numbers.

The whole point is, as I said, that we should not get our knickers all in a twist over mythically-good, internet-rumored magic parts without first doing the numbers. Does resistor noise matter? Well, sometimes yes, and sometimes no. It depends on the circuits and the intended application. Actually being able to calculate or at least estimate the numbers is crucial to not being a slave to "I read it somewhere, long ago and far away, posted by someone who may or may not know".

We still - even in this thread - get admonished about all the magical Golden Age amps being magically better, and by logically faulty extension that their parts are somehow the only ones that can used for good results. I very, very much liked Martin's search for CC resistor distortion. It was real, and used actually measured numbers.

As a quick estimate, the ripple on a 400V-ish power supply can be around 40-50V when the amp is at full roar. Just as a quickie, 40V is about 20db less than 400V, even though we're miscomparing DC with AC; bear with me here. So the noise voltage of a 10K resistor, being about -170db compared to 400V, would be about -150db compared to 40V of ripple. That's a ratio of ten to the -7.5 power, which is about 31 one-billionths of the ripple voltage.

Or, even better, what would the S/N ratio be with a 100k resistor on the plate of a triode? That seems like some math that would be pretty appropriate considering the source you linked is primarily focused on the passive intermodulation (PIM) that distorts an actual signal by intermodulating it with noise.

Now you're on it! The question is what are the circuit-appropriate numbers? Good on you.
The noise of a triode input stage was studied to death between about 1920 and 1980. The noise of the triode stage at the output plate is computed by RMS-adding the noise of the input resistances (often a 220k to 1M resistance) times the stage's voltage gain, the triode's internal noise sources, the noise of the plate resistor as attenuated by the triode internal plate resistance and any loading and attenuation of the plate signal by the next stage input. Once again, you have to consider the whole circuit, not just one part.

The thermal noise of a 100K resistor at 25C (... thermal, goes up with temperature) over a 10kHz bandwidth (noise is calculated within some bandwidth) is 4.06 microvolts, which is about -108db referred to 1Vrms. The noise of a 1M grid leak resistor at 25C, 10kHz bandwidth, is 12.8 microvolts, -97.8dbv; call it 10db more than the 100K plate resistor. The grid leak resistor's noise is then multiplied by the gain of the stage, 23 to 26db, to make the thermal noise contribution of a 1M input resistor be about -74.8dbv to -71.8dbv at the plate; this is about 36db worse than the 100K plate resistor's contribution. The answer for the100K plate resistor is - don't sweat the 100K resistor noise, worry a LOT about the input and first stage triode noise long before you worry about the plate resistor.

The second lesson here is that if all the following stages are similar to the input stage, you can largely ignore their noise contribution because the input stage noise is multiplied by all the following gains.

These approximations roll out so easily because I learned them from some great sources. I highly, highly recommend the Radiotron Designer's Handbook, the AudioCyclopedia, and Vacuum Tube Amplifiers by the MIT Radiation Lab. I have hardcopy of each of these, and protect them at least as well as my credit cards.

[martin manning]

Here's a nice measurement of resistor noise that confirms published ranking of different resistor construction types for both TH and SMT: https://www.beis.de/Elektronik/Resistor ... Noise.html
If you want low noise, choose low noise resistors, especially for low s/n areas and where the circuit will be prone to generating noise.

In audio circles it's definitely a fringe element that talks about the sound qualities of different resistors other than their noise level, and there are market forces around some basic human desires at play in that area. I'm definitely in the "it's just noise" camp, but I'm open to the idea that the presence or absence of noise in the audio may alter the perception of other frequencies and harmonics.

[R.G.]

Fwiw Koa Speer and yageo still produce higher wattage cf in many values

For whatever it's worth, CCs are not just CCs. I suspect that modern-manufacture CCs have had their binders changed since the Golden Age. Most of the techie stuff I've read about what makes CC especially good or bad attributes the good or bad to the binder, not the carbon granules so much. A discussion of CC versus other types probably needs to include some discussion of the binders.

AAAACK! THE BINDERS!!??!! Who knows what the Golden Age binders were, what the modern binders are and are not, and what that change does compared to the decades-old binders in decades-old CCs. Were Golden-Age CCs as good when new as we think they are now? We know the Golden Age resistors have changed by aging over the years, because drifted resistors are a very, very common issue with repair techs.

This leads to the question - how valuable is the advice to use or not use CCs in the face of all this quicksand? Is there some way to actually tell whether new CCs are as tone magic as older CCs. Or better? Is older better, in which case we might start aging new CCs before use, or is modern CC better, and how?

[WhopperPlate]

Fwiw Koa Speer and yageo still produce higher wattage cf in many values

For whatever it's worth, CCs are not just CCs. I suspect that modern-manufacture CCs have had their binders changed since the Golden Age. Most of the techie stuff I've read about what makes CC especially good or bad attributes the good or bad to the binder, not the carbon granules so much. A discussion of CC versus other types probably needs to include some discussion of the binders.

AAAACK! THE BINDERS!!??!! Who knows what the Golden Age binders were, what the modern binders are and are not, and what that change does compared to the decades-old binders in decades-old CCs. Were Golden-Age CCs as good when new as we think they are now? We know the Golden Age resistors have changed by aging over the years, because drifted resistors are a very, very common issue with repair techs.

This leads to the question - how valuable is the advice to use or not use CCs in the face of all this quicksand? Is there some way to actually tell whether new CCs are as tone magic as older CCs. Or better? Is older better, in which case we might start aging new CCs before use, or is modern CC better, and how?

Great good for thought! I have heard this argument with modern Cc . I honestly don’t care for Cc within the signal path for most things , so I never nerded out down that rabbit hole . I just use whatever nos I can find for the b+. Obviously not very sustainable , but it is what it is

Fwiw the Koa and yageo I referenced are CF , not cc

[WhopperPlate]

In audio circles it's definitely a fringe element that talks about the sound qualities of different resistors other than their noise level, and there are market forces around some basic human desires at play in that area. I'm definitely in the "it's just noise" camp, but I'm open to the idea that the presence or absence of noise in the audio may alter the perception of other frequencies and harmonics.

Amazing the extent of the potential variability imparted by simply noise … fascinating